Okay, yeah, I think my talk is actually a bit of a bridge between the software pitches

and software development best practices topics.

So I'm going to discuss how does this work now?

Yes, okay.

So my talk is about CASA.

I have a bit of what to describe the package, software package, and then I have a few slides

on the development environment that we use and practices that we practice.

So CASA, the idea is that it's a general purpose radio astronomy data processing package.

In principle, it should be able to process both single dish and radio interferometer

data.

It is the primary package that's used for the ALMA telescope and for the VLA telescope

in the US, but it is used by other telescopes as well.

We've seen that Meerkat uses components of CASA in their data reduction pipeline.

Still unclear what the SKA is going to do.

It's used by both end users and instrument pipelines, which is, I think, an interesting

aspect.

Which also means that there are still many people that simply use it on their own laptop.

But there's also the more pipeline oriented users use it on small clusters.

It's distributed under the LGPL license, so it's fully open source, and it's developed

by a consortium that's led by the National Radio Astronomy Observatory in the US.

But includes partners from Europe, Australia, and Asia.

A little bit of history.

The software started out under the name H++, which really was supposed to be a re-implementation

of the classic APES data reduction package that was written in the early 80s, I believe,

most of the code in Fortran.

This was early 90s, and the idea was to do this in a modern software environment, C++,

which was still very young at that time.

At that point, it also included a data manipulation language called Glitch.

Those high energy physics people who remember the SSC might recognize it.

Glitch was supposed to be part of the software for the superconducting supercollider, as

that SSC stands for.

Unfortunately, that project got cancelled, and so the support for Glitch dropped, and

the astronomy community or the radio astronomy community in itself did not have the resources

to maintain that.

That was replaced in the early 2000s by Python, and it used IPython as the user interface

to the code.

H++ also got a little bit of a bad rap in those days, because it had the reputation

of being slow at not really being useful for anything.

So it also got rebranded.

That's what you do if you've got a bad reputation.

You just rename yourself and make a fresh start.

Basically the functionality that Glitch would do, as in having sort of a higher level, easy

for users to use data manipulation language, got replaced with Python.

I think that has paid off very well, because the rest of astronomy has basically moved

to Python for that purpose as well.

NumPy and SciPy have really established themselves in the last 10 years, I think.

Still there was the impression that CASA itself was a bit of a monolithic application, but

wasn't very flexible.

Not all the partners in the consortium wanted to continue developing it.

So that led to a split of the code, where the more basic fundamental algorithms got